Building Power Demand Forecasting Using Machine Learning: Application for an Office Building in Danang

Toward 2030, 20% of existing buildings and all new construction would have to be zero-carbon ready for reaching the global goal of carbon neutrality by 2050. As a member state of UNFCCC, Vietnam has come up with mechanisms and policies to ensure compliance with strong commitments at COP26. However, energy efficiency design strategies for new construction would not guaranty the energy reduction if buildings were not appropriately operated. To contribute the objective of energy savings and lowering the building’s operational costs, this paper presents a Machine Learning approach for an office building, located in Danang, to forecast the power demand based on time, weather conditions and historical energy consumption. For discovering energy consumption pattern and correlation between features, exploratory data analysis has been performed to initially investigate data sets and summarize their main characteristics. After selecting features that have strong relationship with the target, the power demand of this office building has been predicted using different Machine Learning methods. Forecasting results will be compared to define the relevant method.